We'll learn a lot more about this topic as the semester goes along. So, this introduction will be brief and will stick to some definitions.
Here's the core point: We tend to think about communication in English as involving a process of selecting individual words and putting those words together in the correct order to make sentences. That idea about how English works is not quite accurate. In fact, it is wrong in ways that are extremely important for us to understand as language teachers.
First, all language use is local. We select the words and grammar that we use based on the contexts in which we will use them. If we want to tell a story about the past, we select past tense and past time adverbs along with vocabulary appropriate to the story; and we tend to use indirect and direct quotation and proper nouns and pronouns to refer to those proper nouns; and chronological order and adverbs to signal that order.
Second, words tend to stick together; they hand around together in packs. Those packs are based on meaning...so there are sets of words to use to talk about particular topics. That is, we tend to be working with sets of words rather than single words. These sets can be anticipated by us...sets of words to use to talk about families or work or politics. But some word relationships are almost impossible to anticipate and these are the word relationships that linguists label collocations and lexical bundles.
Collocation is the name for pairs of words that tend to occur together in discourse. When we analyze the vocabulary in large sets of English in a corpus, we find sets of words that show up together. The tendency is that when you find one of the words, you'll fine the other word in that same setting.
On page 18 in the Longman Grammar, we're given the example of differences between two words: broad and wide. They are a lot alike in meaning but they tend to be used with different nouns. The examples in the book are broad accent, broad agreement, broad daylight, broad grin, broad shoulders and wide appeal, wide area, wide experience, wide interests, wide margin. We talk about this relationship by saying that "broad collocates with accent." Or, we say that "accent, agreement, daylight, grin, and shoulders are collocates of broad."
These relationships are real. They exist. They are fundamental to the structure of English and to communication in English. But, they are not relationships that we can come up with just sitting around thinking about English vocabulary. Collocation analysis requires corpora and computers and software to analyze the vocabulary relationships in the corpora. We are now getting dictionaries that provide us with collocational information about words.
This topic also ties to the issue of the use of "authentic" materials for teaching ESL/EFL: there's always a chance that if we write materials to illustrate a grammar point that we'll violate the rules of English vocabulary and come up with examples that teach students the wrong collocations.
On page 455, in the Glossary of Terms, collocation has this definition: "a combination of lexical words which frequently co-occur in texts: little + baby, small + amount, make + (a) + mistake."
We are also learning that words very often come in sets of 3 or more words that are just about always just together. We do not take one word and combine it freely with the other words. These words must be used together. These include sets such as the following:
Research has shown that one of the most frequent sets of words in English conversation is "I don't want to...." Isn't that just lovely and so true to human nature?!
For now, just notice 2 things about these "lexical bundles": 1. certain words occur together as a set and 2. that set is open to addition of other words to complete the meaning and the grammar.